An inequality for the group chromatic number of a graph



Similar documents
Total colorings of planar graphs with small maximum degree

SHARP BOUNDS FOR THE SUM OF THE SQUARES OF THE DEGREES OF A GRAPH

Labeling outerplanar graphs with maximum degree three

Mean Ramsey-Turán numbers

Cycles in a Graph Whose Lengths Differ by One or Two

Odd induced subgraphs in graphs of maximum degree three

Connectivity and cuts

Introduction to Graph Theory

P. Jeyanthi and N. Angel Benseera

SCORE SETS IN ORIENTED GRAPHS

The positive minimum degree game on sparse graphs

Every tree contains a large induced subgraph with all degrees odd

A 2-factor in which each cycle has long length in claw-free graphs

Zachary Monaco Georgia College Olympic Coloring: Go For The Gold

On Some Vertex Degree Based Graph Invariants

All trees contain a large induced subgraph having all degrees 1 (mod k)

BOUNDARY EDGE DOMINATION IN GRAPHS

Cacti with minimum, second-minimum, and third-minimum Kirchhoff indices

SEQUENCES OF MAXIMAL DEGREE VERTICES IN GRAPHS. Nickolay Khadzhiivanov, Nedyalko Nenov

UPPER BOUNDS ON THE L(2, 1)-LABELING NUMBER OF GRAPHS WITH MAXIMUM DEGREE

COMBINATORIAL PROPERTIES OF THE HIGMAN-SIMS GRAPH. 1. Introduction

On the maximum average degree and the incidence chromatic number of a graph

A simple criterion on degree sequences of graphs

Cycle transversals in bounded degree graphs

I. GROUPS: BASIC DEFINITIONS AND EXAMPLES

On Integer Additive Set-Indexers of Graphs

The chromatic spectrum of mixed hypergraphs

A REMARK ON ALMOST MOORE DIGRAPHS OF DEGREE THREE. 1. Introduction and Preliminaries

A characterization of trace zero symmetric nonnegative 5x5 matrices

Large induced subgraphs with all degrees odd

Graphs without proper subgraphs of minimum degree 3 and short cycles

On the independence number of graphs with maximum degree 3

136 CHAPTER 4. INDUCTION, GRAPHS AND TREES

8.1 Min Degree Spanning Tree

A threshold for the Maker-Breaker clique game

Non-Separable Detachments of Graphs

Tools for parsimonious edge-colouring of graphs with maximum degree three. J.L. Fouquet and J.M. Vanherpe. Rapport n o RR

Clique coloring B 1 -EPG graphs

School of Industrial and System Engineering Georgia Institute of Technology Atlanta, Georgia 30332

Mathematics Course 111: Algebra I Part IV: Vector Spaces

8. Matchings and Factors

On the k-path cover problem for cacti

A Turán Type Problem Concerning the Powers of the Degrees of a Graph

3. Eulerian and Hamiltonian Graphs

Ph.D. Thesis. Judit Nagy-György. Supervisor: Péter Hajnal Associate Professor

Cycles and clique-minors in expanders

CS 598CSC: Combinatorial Optimization Lecture date: 2/4/2010

F. ABTAHI and M. ZARRIN. (Communicated by J. Goldstein)

ON GENERALIZED RELATIVE COMMUTATIVITY DEGREE OF A FINITE GROUP. A. K. Das and R. K. Nath

Strong Ramsey Games: Drawing on an infinite board

Discrete Mathematics Problems


Triangle deletion. Ernie Croot. February 3, 2010

Discrete Mathematics & Mathematical Reasoning Chapter 10: Graphs

Solutions for Practice problems on proofs

Finding and counting given length cycles

Every Positive Integer is the Sum of Four Squares! (and other exciting problems)

Handout #Ch7 San Skulrattanakulchai Gustavus Adolphus College Dec 6, Chapter 7: Digraphs

The Goldberg Rao Algorithm for the Maximum Flow Problem

On parsimonious edge-colouring of graphs with maximum degree three

FALSE ALARMS IN FAULT-TOLERANT DOMINATING SETS IN GRAPHS. Mateusz Nikodem

Midterm Practice Problems

INCIDENCE-BETWEENNESS GEOMETRY

MATH10212 Linear Algebra. Systems of Linear Equations. Definition. An n-dimensional vector is a row or a column of n numbers (or letters): a 1.

Classification of Cartan matrices

Generalized Induced Factor Problems

Extremal Wiener Index of Trees with All Degrees Odd

Definition Given a graph G on n vertices, we define the following quantities:

A NOTE ON OFF-DIAGONAL SMALL ON-LINE RAMSEY NUMBERS FOR PATHS

Chapter 3. Cartesian Products and Relations. 3.1 Cartesian Products

Sample Induction Proofs

THE DIMENSION OF A VECTOR SPACE

Single machine parallel batch scheduling with unbounded capacity

On end degrees and infinite cycles in locally finite graphs

n k=1 k=0 1/k! = e. Example 6.4. The series 1/k 2 converges in R. Indeed, if s n = n then k=1 1/k, then s 2n s n = 1 n

Math 319 Problem Set #3 Solution 21 February 2002

DATA ANALYSIS II. Matrix Algorithms

x a x 2 (1 + x 2 ) n.

Continued Fractions and the Euclidean Algorithm

Page 331, 38.4 Suppose a is a positive integer and p is a prime. Prove that p a if and only if the prime factorization of a contains p.

Full and Complete Binary Trees

Embedding nearly-spanning bounded degree trees

Degree Hypergroupoids Associated with Hypergraphs

Row Ideals and Fibers of Morphisms

INDISTINGUISHABILITY OF ABSOLUTELY CONTINUOUS AND SINGULAR DISTRIBUTIONS

GREATEST COMMON DIVISOR

ON INDUCED SUBGRAPHS WITH ALL DEGREES ODD. 1. Introduction

Conductance, the Normalized Laplacian, and Cheeger s Inequality

Transcription:

Discrete Mathematics 307 (2007) 3076 3080 www.elsevier.com/locate/disc Note An inequality for the group chromatic number of a graph Hong-Jian Lai a, Xiangwen Li b,,1, Gexin Yu c a Department of Mathematics, West Virginia University, Morgantown, WV 26505, USA b Department of Mathematics, Huazhong Normal University, Wuhan 430079, China c Department of Mathematics, University of Illinois at Urbana-Champaign, Urbana, IL 61801, USA Received 29 November 2004; received in revised form 8 December 2006; accepted 13 March 2007 Available online 15 March 2007 Abstract We give an inequality for the group chromatic number of a graph as an extension of Brooks Theorem. Moreover, we obtain a structural theorem for graphs satisfying the equality and discuss applications of the theorem. 2007 Elsevier B.V. All rights reserved. Keywords: Brooks Theorem; Group coloring; Group chromatic number 1. Introduction Let δ(g), Δ(G) and χ(g) denote the minimum degree, the maximum degree and the chromatic number of G, respectively. A graph G is unicyclic if G contains only one cycle. Brooks Theorem [1] is well known as follows. Theorem 1.1. If G is a connected graph, then χ(g) Δ(G) + 1 with equality if and only if G is complete or an odd cycle. Let ρ(g) be the largest eigenvalue of the vertex-adjacency matrix of G, i.e., the matrix whose i, j entry is 1 if vertices i and j are connected and 0 otherwise. Wilf [6] improved Brooks Theorem by replacing Δ(G) by ρ(g) and proved that χ(g) ρ(g) + 1 with equality if and only if G is an odd cycle or a complete graph. Let γ(g) be a real function on a graph G satisfying the following two properties: (P1) If H is an induced subgraph of G, then γ(h ) γ(g). (P2) γ(g) δ(g) with equality if and only if G is regular. Szekeres and Wilf [5] extended Theorem 1.1 by replacing Δ(G) by γ(g), as follows. Corresponding author. E-mail address: xwli2808@yahoo.com (X. Li). 1 Supported partially by the Natural Science Foundation of China (10571071) and by SRF for ROCS, State Education Ministry of China. 0012-365X/$ - see front matter 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.disc.2007.03.014

H.-J. Lai et al. / Discrete Mathematics 307 (2007) 3076 3080 3077 Theorem 1.2. If γ(g) is a real function on a graph G with properties (P1) and (P2), then χ(g) γ(g) + 1. Cao [2] observed Szekeres and Wilf s result and proved that the equality holds if and only if G is an odd cycle or a complete graph. Unfortunately, there is a minor error in his result in the sense that there exists a counterexample as shown below. For a connected graph G, we define 0 if G = K 1, 1 if G = K γ(g) = 2, 2 if G is unicyclic or a tree and G = K 2, Δ(G) otherwise. For a disconnected graph G, we define γ(g) = max{γ(g i ) : G i is a connected component of G}. It follows that γ(g) satisfies (P1) and (P2). Therefore, when G is a connected unicyclic graph with an odd girth and G is not isomorphic to a cycle, we have χ(g) = γ(g) + 1, but G is neither a cycle nor a complete graph. One asks when the equality in Theorem 1.2 holds. Moreover, when the equality holds for a graph G, what kind of graph G is. In this note, we will discuss these questions in group coloring. One easily obtains similar results in coloring. We end this section with some important terminology. Graphs in this note are finite and simple. Unless otherwise stated, notations and terminology in graph theory are standard. Denote by H G the fact that H is an induced subgraph of G and by N(v) the set of neighbors of v in G. The concept of group coloring was first defined in [3]. Let G be a graph and A anabelian group. Denote by F(G,A)the set of all functions from E(G) to A and by D an orientation of E(G).Forf F(G,A),an(A, f )-coloring of G under the orientation D is a function c : V (G) A such that for every directed edge uv from u to v, c(u) c(v) = f (uv). G is A-colorable under the orientation D if for any function f F(G,A), G has an (A, f )-coloring. Jaeger et al. [3] proved that A-colorability is independent of the choice of the orientation. The group chromatic number of a graph G, denoted by χ g (G), is defined to be the minimum m for which G is A-colorable for any Abelian group A of order at least m. It follows that for any graph G, χ(g) χ g (G). 2. Main results The group coloring version of Brooks Theorem was proved in [4], as follows. Theorem 2.1. For any connected graph G, χ g (G) Δ(G) + 1, where equality holds if and only if Δ(G) = 2 and G is a cycle; or Δ(G) 3 and G is complete. Theorem 2.1 tells us that when Δ(G) = 2, the equality holds if and only if G is a cycle, which is different from Theorem 1.1. Motivated by Theorem 2.1, the authors first consider to generalize this extension of Brooks Theorem. For a graph G, we now introduce the concept of χ g (G)(or χ(g))-semi-critical graphs as follows. A graph G is defined to be χ g (G)(or χ(g))-semi-critical if χ g (G v) < χ g (G)(or χ(g v) < χ(g)), for every vertex v G with d(v) = δ(g). It is easy to see that a cycle is semi-critical and so is a complete graph. For a graph G, there exists a subgraph H such that H is a χ g (H )-semi-critical subgraph. Moreover, χ g (G) = χ g (H ). χ g (G)-semi-critical graph has the following properties which play a key role in the proof of Theorem 2.4. Lemma 2.2. LetGbeagraph, v V (G), and H = G v. If d G (v) < χ g (H ), then χ g (G) = χ g (H ). Proof. We assume, without loss of generality, that G is oriented such that all the edges incident with v are oriented from v. Let A be an Abelian group with A χ g (H ) and f F(G,A). It follows that for f H : E(H) A, there is a

3078 H.-J. Lai et al. / Discrete Mathematics 307 (2007) 3076 3080 function c : V(H) A such that c(x) c(y) = f(xy)for each directed edge from x to y. Let N(v)={v 1,v 2,...,v l }, where l = d(v). Since l<χ g (H ) A, there is an a A l i=1 {c(v i ) + f(vv i )}. Define c 1 : V (G) A by { c(u) if u V(H), c 1 (u) = a if u = v, which implies that G is A-colorable and hence χ g (G) χ g (H ). We conclude our lemma by inequality χ g (H ) χ g (G) since H G. Lemma 2.3. IfGisχ g (G)-semi-critical with χ g (G) = k, then d G (v) k 1 for all v V (G). Proof. Suppose, to the contrary, that G has a vertex v 0 with d G (v 0 ) = δ(g) k 2. Since G is a χ g -semi-critical graph with χ g (G) = k, wehaveχ g (G v 0 ) k 1. If χ g (G v 0 ) = k 1, by Lemma 2.2, χ g (G) = χ g (G v 0 ) = k 1, contrary to that χ g (G) = k. We thus assume that χ g (G v 0 ) k 2. By similar argument as in the proof of Lemma 2.2, we obtain that χ g (G) k 1. This contradiction proves our lemma. Theorem 2.4. If G is a connected graph and γ(g) is a real function satisfying (P1) and (P2), then χ g (G) γ(g) + 1. Moreover, ifgisχ g (G)-semi-critical, then χ g (G) = γ(g) + 1 if and only if G is a cycle or a complete graph. Proof. Let k = χ g (G) and let H G be a k-semi-critical induced subgraph. By (P1), γ(h ) γ(g). By Lemma 2.3, δ(h ) k 1. By (P2), k 1 δ(h ) γ(h ) γ(g), which implies that χ g (G) = k γ(g) + 1. If G is a cycle C or a complete graph K n, by (P2), γ(c) = 2 and γ(k n ) = n 1. Our sufficiency follows from that χ g (C) = 3 and χ g (K n ) = n (see [4]). Conversely, let G be a χ g (G)-semi-critical graph such that χ g (G) = γ(g) + 1. By Lemma 2.3 and (P2), χ g (G) 1 δ(g) γ(g) = χ g (G) 1, which implies δ(g) = γ(g) = χ g (G) 1. By (P2), G is regular. Therefore, χ g (G) = γ(g) + 1 = δ(g) + 1 = Δ(G) + 1. By Theorem 2.1, G is a cycle or a complete graph. Theorem 2.4 answers the first question posed in Section 1. For vertex coloring, we easily obtain the similar results as Lemmas 2.2 and 2.3. Corollary 2.5 extends Szekeres and Wilf s result, that is Theorem 1.2. When we prove Corollary 2.5, we only need a slight change by using Theorem 1.1 instead of using Theorem 2.1. Corollary 2.5. If G is a connected graph and γ(g) is a real function satisfying (P1) and (P2), then χ(g) γ(g) + 1. Moreover, if G is semi-critical, then χ(g) = γ(g) + 1 if and only if G is an odd cycle or a complete graph. We are ready to discuss the structure of connected graphs with χ g (G) = γ(g) + 1, where γ(g) is a real function satisfying (P1) and (P2). At first we define a family of graphs as follows. For a positive integer m, define F(m) to be the set of connected simple graphs such that: (1) For m = 1, G F(m) if and only if G = K 1. (2) For m = 2, G F(m) if and only if G = K 2. (3) For m = 3, G F(m) if and only if either G is a cycle, or H = G v for some H F(m) and a vertex v with d(v) = 1. (4) For m 4, G F(m) if and only if either G = K m or H = G v for some H F(m) and a vertex v with d(v) m 2. The goal of the definition is to prove Theorem 2.6 with a real function γ(g). Thus, we define F(1) ={K 1 } and F(2) ={K 2 }. Note that if m = 3, F(3) is the set of connected unicyclic graphs and if m 4, then F(m) is the set of connected graphs such that by the removal of all vertices of degree no more than m 2, the resulting graph is K m. The following theorem answers the second question posed in Section 1. Theorem 2.6. Let G be a connected graph. If χ g (G) = γ(g) + 1, then G F(m) where m = χ g (G). Proof. Let m = χ g (G) and let H 0 be a χ g (G)-semi-critical induced subgraph of G. It is trivial when χ g (G) = 1, 2. We thus assume that χ g (G) 3. It follows that χ g (H 0 ) = χ g (G) = γ(g) + 1 γ(h 0 ) + 1 χ g (H 0 ). It implies that

H.-J. Lai et al. / Discrete Mathematics 307 (2007) 3076 3080 3079 χ g (H 0 ) = γ(h 0 ) + 1 and γ(h 0 ) = γ(g). By Theorem 2.4, H 0 is a cycle or a complete graph and hence δ(h 0 ) = m 1. If δ(g) m 1, then m = χ g (G) = γ(g) + 1 δ(g) + 1 m and hence γ(g) = δ(g) = m 1. It follows that G is regular. Since G is connected, G = H 0. When m = 3, H 0 is a cycle and hence G F(3). When m 4, H 0 is a complete graph and hence G F(m). Assume now that δ(g) m 2. It follows that when m = 3, G cannot be a cycle and that when m 4, G is not a complete graph. By Theorem 2.4, G cannot be χ g (G) semi-critical. Therefore, there exists v V (G) with d(v)= δ(g) such that χ g (G) = χ g (G v). We prove that G F(m) by induction on V (G). If V (G) = V(H 0 ) +1, then G v = H 0 F(m). Bythe definition of F(m), G F(m). Assume now that V (G) > V(H 0 ) +1. By Lemma 2.2 and by Property (P1), χ g (G v) = χ g (G) = γ(g) + 1 γ(g v) + 1 χ g (G v). It implies that χ g (G v) = γ(g v) + 1. We apply the induction hypothesis to G v, G v F(m). Bythe definition of F(m), G F(m). Note that χ(g) χ g (G) γ(g)+1. χ(g)=γ(g)+1 implies χ g (G)=γ(G)+1. When χ(g)=γ(g)+1, G F(m). If G is a unicyclic graph, the cycle cannot be even. Thus, the following corollary is straightforward. Corollary 2.7. Let G be a connected graph and γ(g) a real function satisfying (P1) and (P2). If χ(g) = γ(g) + 1, then G F(m) for some m 3 or G is a unicyclic graph with odd girth or G = K 2 or G = K 1. As applications of Theorem 2.6, we prove Corollaries 2.8 and 2.9, respectively. Corollary 2.8 is the group coloring version of a Cao s result [2]. The k-degree of vertex v in a graph G is defined to be the number of walks of length k from v in G. Denote by T (G) the maximum 2-degree of G and by Δ k (G) the maximum k-degree in a graph G which is the maximum row sum of A k, where A is the vertex-adjacency matrix of G. Corollary 2.8. If G is a connected graph, then each of the following holds: (i) χ g (G) T (G) + 1, where equality holds if and only if G is complete or a cycle. (ii) χ g (G) [Δ k (G)] 1/k + 1, where equality holds if and only if G is complete or a cycle. Proof. (i) It is easy to check that T (G) satisfies (P1) and (P2). By Theorem 2.4, χ g (G) T (G) + 1. When χ g (G) = T (G) + 1, by Theorem 2.6, G F(m). The result follows immediately if m = 1, 2. If m = 3, G is a unicyclic graph and hence χ g (G) = 3. Since T (G) 2 + 2 + x with T (G) = 2 if and only if x = 0, that is, G must be a cycle. Similarly, when m 4, G F(m) and T (G) = m 1 and hence G is a complete graph. Conversely, if G is a complete graph or a cycle, then G is semi-critical. By Theorem 2.4, χ g (G) = T (G) + 1. (ii) The proof is similar. For a graph G, the average degree and the maximum average degree of G are defined to be l(g) = 2 E(G) / V (G) and L(G) = max{l(h) : H is an induced subgraph of G}, respectively. Clearly, L(G) satisfies (P1) and (P2). Therefore, the following corollary is straightforward. Corollary 2.9. If G is a connected graph with maximum average degree L(G), then χ g (G) L(G)+1, where equality holds if and only if G F(m), where m = χ g (G). Acknowledgment The authors would like to thank the referees for the valuable comments which improve the presentation. References [1] R. Brooks, On colouring the nodes of a network, Proc. Cambridge Philos. Soc. 37 (1941) 194 197. [2] D. Cao, Bounds on eigenvalues and chromatic number, Linear Algebra Appl. 270 (1998) 1 13.

3080 H.-J. Lai et al. / Discrete Mathematics 307 (2007) 3076 3080 [3] F. Jaeger, N. Linial, C. Payan, M. Tarsi, Graph connectivity of graphs a nonhomogeneous analogue of nowhere-zero flow properties, J. Combin. Theory B 56 (1992) 165 182. [4] H.-J Lai, X. Zhang, Group colorability of graphs, Ars Combin. 62 (2002) 299 317. [5] G. Szekeres, H.S. Wilf, An inequality for the chromatic number of a graph, J. Combin. Theory 4 (1968) 1 3. [6] H.S. Wilf, The eigenvalues of a graph and its chromatic number, J. London Math. Soc. 42 (1967) 330 332.